Re: [sipcore] #8: It's not clear when a Proxy should/should-not add H-I for "internal" stuff

I agree that this should be clarified.  We likely need to add
appropriate normative text in the Index section on the calculation of
the index for internal retargeting OR refer explicitly to the
appropriate section in terms of how the indices are calculated.  The
best approach IMHO is that it looks like the 302 scenario - i.e., the
second example below.  Certainly, you won't be able to tell that this
was internal to the proxy by the indices, but it does reflect the
request was retargeted to another logical entity within the same
domain.

Mary.

On Thu, Sep 2, 2010 at 4:11 PM, Worley, Dale R (Dale) <dworley@avaya.com> wrote:
> ________________________________________
> From: Hadriel Kaplan [HKaplan@acmepacket.com]
>
> But they're NOT "Request-URIs".  Not yet.  They're abstract target URIs during the internal local policy stages of routing, until such point as the request gets forwarded.  They're just memory blocks in a system, getting munged based on transformation rules, lookup tables, or possibly non-SIP database queries.
>
> For example, a common routing local policy procedure of originating proxies in North America is to perform the following for received request-URIs identified as phone numbers:
> 1) Normalize the number by removing visual separators, and possibly transform the number to an E.164, by adding a country-code, area-code, leading-plus, etc.
> 2) Do a number portability dip for the called number.  That changes the target URI to include a 'npdi' and possibly 'rn' param info.
> 3) Determine whether the portability-corrected target is local or out of region, or if it's out of the local domain altogether (and to which peer).  That can change the target again, for example by changing the domain name, adding trunk-group info, etc.
> 4) Change the Request-URI to be the new target.
> ________________________________________
>
> Hmmm....  In the sipX architecture, each of those steps would be done separately, by the proxy sending the request to a redirect server which would return a 302.  So yes, each stage of processing would be represented by a separate hi-entry.
>
> But I see your point, What if a proxy does it all internally?
>
> I'd like to say that it can write hi-entry's *as if* each step was described separately by a 302 from a redirect server.  Or alternatively, as if the request had a series of passes through proxies that performed each step.  Those alternatives would lead to History-Info somewhat like these (where I've changed the use of the tags a bit):
>
> History-Info:
>        <sip:5556666@carrier1.com>;index=1,
>        <sip:+16175556666@carrier1.com>;index=1.1,
>        <sip:+15552340000@carrier1.com;npdi>;index=1.1.1,
>        <sip:+15552340000@carrier1.com;npdi>;index=1.1.1.1,
>        <sip:+15552340000@carrier2.com;npdi>;index=1.1.1.1.1
>
> History-Info:
>        <sip:5556666@carrier1.com>;index=1,
>        <sip:+16175556666@carrier1.com>;index=1.1,
>        <sip:+15552340000@carrier1.com;npdi>;index=1.2;from=1.1,
>        <sip:+15552340000@carrier1.com;npdi>;index=1.3;from=1.2,
>        <sip:+15552340000@carrier2.com;npdi>;index=1.4;from=1.3
>
> In general, the recipient of a request containing such a faked H-I wouldn't be able to tell that it had been faked.  However, it would be desirable to have text to explicitly approve such behavior.
>
> Going back to the text of the document, I see how I became misled.  The text in section 5.1 is:
>
>   Retargeting is an iterative process, i.e., a proxy may redirect
>   "internally " more than one time.  A typical example would be a proxy
>   that redirects a request first to a different user (i.e., it maps to
>   a different AOR), and then forwards to a registered contact bound to
>   that new AOR.  A proxy that uses such mechanism SHOULD add multiple
>   hi-entry fields (e.g., bob@example.com to office@example.com to
>   office@192.0.2.5) to provide a logical description of the retargeting
>   process.  A Reason MAY be associated with the hi-targeted-to-uri that
>   has been retargeted as shown in the example in Appendix B.1.
>
> Given that the word "internally" is not defined in any way, and that the example described would be implemented in sipX through multiple redirect server requests, I did not realize that the process described was performed entirely within one SIP element.  This suggests that the text could stand to be improved.
>
> Dale
> _______________________________________________
> sipcore mailing list
> sipcore@ietf.org
> https://www.ietf.org/mailman/listinfo/sipcore
>